Reliability-Oriented Uplink Resource Management in Ultra-Low Latency Mobile Networks

The forthcoming 6G has necessitated more stringent requirements for network latency communication (0.1-1 ms). For this purpose, the Ultra-low Latency Mobile Network (ULMN) integrates open-loop communication, ultra-low latency mobile network association and the cell-free structure towards the minimum end-to-end latency, thus meeting the 6G network latency requirements. However, the ULMN cancels control signaling interactions and traditional centralized resource allocation mode, so both the sender and receiver cannot obtain real-time and accurate channel state information (CSI) in uplink and downlink transmission, which cannot guide reasonable resource allocation and seriously damage network reliability. Uplink communication is more challenging than downlink communication, because a large number of users will make transmission decisions autonomously and blindly to choose channels with poor quality and the same resources to transmit data, resulting in resource conflicts. Therefore, this paper focuses on machine-initiated uplink transmission in ULMN and exploits a reliability-guaranteed distributed resource management scheme. Specifically, we design an novel Multi-Agent Deep Deterministic Policy Gradient based Uplink Resource Management (MADDPG-URM) algorithm to maximize the average probability of success per transmission over time, where senders use historical transmission experience and additional sensor information for resource occupation, association decision-making and power control in a distributed way. The simulation results show that the proposed MADDPG-URM algorithm is more suitable for high network load than radio resource utilization information generation strategies with AN assistance (AA-RRUI) and random algorithm.